Other Methodological Issues

Many of the methodological issues involved in the study of infant mortality are similar, if not identical, to those associated with research in other substantive areas and are more appropriately discussed in the demographic methods literature. However, certain methodological matters need to be addressed here because they are integral to the success of efforts to describe and/or account for variation in infant mortality.

distributional issues. The birth weight distribution ''is composed of at least two statistically identifiable subpopulations'' (Gage 2000: 181). The first, sometimes denoted the predominant distribution (Wilcox and Russell 1986), is essentially normal (Gaussian) and consists of most infants born at normal weight and term gestation and among whom the infant mortality rate is low. The second is the ''residual distribution of small births,'' among whom far and away the largest proportion of deaths in the first year occurs (Wilcox and Russell 1986).7 A useful way of dealing with this heterogeneous distribution is through models of weight-specific perinatal mortality (Wilcox and Russell 1983). Wilcox and Russell have demonstrated that meaningful comparisons of the infant mortality rate across populations8 can be achieved ''by plotting each weight-specific mortality curve relative to its own birthweight distribution'' (Wilcox and

7 Actually, a third component of the distribution might be identified, namely, extremely heavy (macrosomic) births which are characterized by an upward inflection in mortality risk (Wilcox and Russell 1983).

8 The Wilcox and Russell research also has implications for modeling the individual risk of infant death (Solis, Pullum, and Frisbie 2000). At this juncture, however, the focus is entirely on comparisons of rates across populations.

Russell 1986: 188). Failure to take these insights into account leads to a ''low birth weight paradox,'' viz., puzzlement over the finding that ''LBW babies in high-risk populations . . . usually have lower mortality than LBW babies in better-off populations'' (Wilcox 2001a: 1234). A prime example occurs in what at first glance seems to be an enigmatic effect of smoking. While it early became clear that both low birth weight and infant mortality are more likely among babies of mothers who smoke, it was also discovered that ''LBW babies born to mothers who smoked had lower infant mortality than the LBW babies of mothers who did not smoke'' (Wilcox 2001a: 1234; emphasis in the original). The puzzle is solved when infant mortality is plotted by relative birth weight (adjusted to z-scores), in which case ''[m]ortality with mother's smoking is higher across the whole range of weights'' (Wilcox 2001a: 1237).

Another approach to analyzing the relationship between the primary (predominant) and secondary (residual) components of the birth weight distribution involves the use of ''mixture models,'' as applied by Gage (2000) and Gage and Therriault (1998), in which the two components are taken as Gaussian, but with different means and standard deviations. One of the most important results emerging from mixture models is that both birth weight and gestational age distributions ''vary significantly between the sexes and among ethnic groups'' (Gage 2000: 181).

POPULATION-SPECIFIC STANDARDS. Findings of heterogeneity in pregnancy outcome distributions by race/ethnicity and sex support the contention by Alexander and colleagues that ''[g]iven that the general norms for preterm, postterm, and fetal growth measures may be largely derived from White populations, more information is needed to assess whether or not the ongoing use of these one-size-fits-all standards may result, for some ethnic groups, in invalid risk assessments and the misidentification of infants in need of intervention services'' (1999a: 77). This argument is consonant with the conclusions by Kline and colleagues who note that ''measures of development do not correspond with post-conception age in a way that is consistent (across populations)____Sex is one such criterion____Ethnicity or race is another'' (1989: 188), as well as with the call by Wilcox and Russell (1990) for population-specific standards.

5 Easy Ways To Stop Smoking

5 Easy Ways To Stop Smoking

Your first day without cigarettes can be difficult, but having a plan will make it easier! Learn what steps to take on the day you quit smoking.

Get My Free Ebook

Post a comment